Dataset Viewer
Auto-converted to Parquet Duplicate
image
image
label
class label
text
string
pair_id
int32
0real
ns2.dns.nic.swiftcover
1
1phishx
ns2.dns.nic.swiftcovér
1
0real
ns4.dns.nic.swiftcover
2
1phishx
ns4.dns.nic.swiƒtcover
2
0real
anDY-YOU-GOOnie.swiftcover
5
1phishx
anDY-YOU-GOOnie.ṡwiftcover
5
0real
xn--D1ACAM7Agia1k.swiftcover
6
1phishx
xn--D1ACAM7Agia1k.swiftᴄover
6
0real
ns1.dns.nic.swiftcover
7
1phishx
ńs1.dns.nic.swiftcover
7
0real
whATAREYOUDOinghereanyway.swiftcover
8
1phishx
whATAREYOUDOingheṛeanyway.swiftcover
8
0real
ns5.dns.nic.swiftcover
9
1phishx
nʂ5.dns.nic.swiftcover
9
0real
ns3.dns.nic.swiftcover
10
1phishx
ns3.dns.nic.sⱳiftcover
10
0real
weFIXLAMECOMputers.swiftcover
12
1phishx
ᴡeFIXLAMECOMputers.swiftcover
12
0real
anASTASI-KAY.swiftcover
14
1phishx
anASTASI-KAY.swiftcovęr
14
0real
blog.adac
17
1phishx
blog.adaᴄ
17
0real
nic.adac
18
1phishx
nic.âdac
18
0real
b.nic.adac
20
1phishx
b.nić.adac
20
0real
an-TOIne-7.adac
21
1phishx
am-TOIne-7.adac
21
0real
a.nic.adac
22
1phishx
a.niƈ.adac
22
0real
amIMOBile.adac
23
1phishx
ǎmIMOBile.adac
23
0real
presse.adac
24
1phishx
presse.aḍac
24
0real
wiTZOO.adac
29
1phishx
wiTZOO.aɖac
29
0real
whois.nic.save
30
1phishx
wẖois.nic.save
30
0real
www.nic.save
31
1phishx
ẉww.nic.save
31
0real
trENNUngsunterhalt.save
33
1phishx
trENNUngsunterhalt.saṽe
33
0real
anDREEapeach.save
35
1phishx
añDREEapeach.save
35
0real
whois.nic.buy
36
1phishx
whoɩs.nic.buy
36
0real
www.nic.buy
38
1phishx
ẅww.nic.buy
38
0real
weEHAwkencatering.buy
39
1phishx
weEHAwkencaterıng.buy
39
0real
wiNDOverapartments.buy
41
1phishx
wiNDOveràpartments.buy
41
0real
chrysler.sn
44
1phishx
chṙysler.sn
44
0real
gsholding.sn
45
1phishx
gsħolding.sn
45
0real
www.hassmar.gouv.sn
47
1phishx
www.haꜱsmar.gouv.sn
47
0real
www.hepatites.sn
48
1phishx
www.hepatǐtes.sn
48
0real
ns2.arc.sn
49
1phishx
ns2.arᴄ.sn
49
0real
www.satcom.sn
50
1phishx
www.satcoᴍ.sn
50
0real
fmpos.ucad.sn
51
1phishx
fmṗos.ucad.sn
51
0real
eisdns1.expertis.sn
52
1phishx
eiṡdns1.expertis.sn
52
0real
solosport.sn
53
1phishx
sołosport.sn
53
0real
sodishop.sn
54
1phishx
sodishoṗ.sn
54
0real
www.socomec.sn
55
1phishx
ẅww.socomec.sn
55
0real
securemail.neurotech.sn
57
1phishx
securemail.ńeurotech.sn
57
0real
mailscan1.3f.sn
59
1phishx
ḿailscan1.3f.sn
59
0real
myecole.sn
61
1phishx
mýecole.sn
61
0real
birdingforalark.blogspot.sn
62
1phishx
ƅirdingforalark.blogspot.sn
62
0real
tobad.sn
63
1phishx
tobåd.sn
63
0real
www.oracle.sn
65
1phishx
www.orácle.sn
65
0real
pub.uadb.edu.sn
66
1phishx
puḇ.uadb.edu.sn
66
0real
www.abb.sn
67
1phishx
ẁww.abb.sn
67
0real
senre.sn
68
1phishx
ŝenre.sn
68
0real
www.dnb.com.sn
69
1phishx
www.dnb.cöm.sn
69
0real
homeview.sn
71
1phishx
homêview.sn
71
0real
ns.univ-thies.sn
74
1phishx
nṣ.univ-thies.sn
74
0real
www.optioncarriere.sn
76
1phishx
www.optioncărriere.sn
76
End of preview. Expand in Data Studio

GlyphNet: Homoglyph Domains Dataset

Data for detecting homoglyph phishing domains (e.g. facebook.com spoofed with visually-similar Unicode characters). Every genuine domain is paired with a synthetically generated homoglyph variant, and each domain is also rendered to a 256x256 grayscale image so the task can be tackled as text or image classification.

Paper: arXiv:2306.10392 · Code: github.com/Akshat4112/Glyphnet

Configs & splits

Both configs share the same train / validation / test split (70/20/10). Splits are assigned per pair (seeded), so a genuine domain and its homoglyph always fall in the same split - preventing leakage between near-identical pairs.

Config Rows Fields
pairs 1,285,579 pairs domain (genuine), homoglyphs (spoofed), pair_id
images 2,571,158 images image (256x256 grayscale PNG), label (real/phish), text, pair_id

pair_id links the two images rows (one real, one phish) that came from the same source pair, and matches the pair_id in the pairs config.

from datasets import load_dataset

pairs  = load_dataset("Akshat4112/Glyphnet", "pairs")
images = load_dataset("Akshat4112/Glyphnet", "images")
train_img = images["train"]        # or "validation" / "test"

How it was generated

Homoglyphs are produced by substituting one or two characters with Unicode confusables (code/dataGeneration.py). Images are rendered with the DejaVu Sans font - Arial (used in the paper) is proprietary and not redistributed here, so glyph shapes may differ slightly from the original figures.

Intended uses & limitations

  • Intended for research on homoglyph / IDN-homograph phishing detection.
  • Homoglyphs are synthetic, not harvested from real attacks, so the distribution may not match live phishing in the wild.
  • The genuine and spoofed strings within a pair are near-identical; always respect the provided splits (or group by pair_id) to avoid leakage.
  • Rendering font differs from the paper (DejaVu Sans vs Arial).

License

MIT (same as the source repository).

Citation

If you use this dataset in your research, please cite the GlyphNet paper:

@article{gupta2023glyphnet,
  title   = {GlyphNet: Homoglyph domains dataset and detection using attention-based Convolutional Neural Networks},
  author  = {Gupta, Akshat and Tomar, Laxman Singh and Garg, Ridhima},
  journal = {arXiv preprint arXiv:2306.10392},
  year    = {2023}
}

Gupta, A., Tomar, L. S., & Garg, R. (2023). GlyphNet: Homoglyph domains dataset and detection using attention-based Convolutional Neural Networks. arXiv:2306.10392.

Downloads last month
96

Paper for Akshat4112/Glyphnet